Page 1
Proceedings of the XVII ECSMGE-2019 Geotechnical Engineering foundation of the future
ISBN 978-9935-9436-1-3 © The authors and IGS: All rights reserved, 2019 doi: 10.32075/17ECSMGE-2019-0233
IGS 1 ECSMGE-2019 - Proceedings
Some observations on the design and construction of
wet soil mixing in the UK Quelques observations sur la conception et la construction de
mélanges de sol humide au Royaume-Uni A. O’Brien
GE Solutions Consulting Ltd, Whitburn, Scotland
ABSTRACT: Mass soil mixing and deep soil mix columns are a versatile ground improvement technology for
marginal and brownfield sites. Dry soil mixing is relatively common in the UK for improvement of ground with
very wet and/or organic material. Wet mixing is less commonplace and involves introduction of a fluid grout
with simultaneous disaggregation of the soil with a rotating mixing tool. This paper presents the results of strength
verification testing carried out across multiple projects in the UK covering differing soil types with varying
project specification criteria. Some conclusions are drawn with regards to the factors affecting strength
progression and in understanding the mechanics of the mixing process. Mixing time per unit volume of mixed
material is identified as an important parameter for mass mixing. In addition, discrete element modelling has
shown promise in understanding the mechanics of deep column mixing.
RÉSUMÉ: Les colonnes de mélange de sol en masse et de mélange de sol en profondeur constituent une
technologie polyvalente d'amélioration du sol pour les terrains marginaux et les sites contaminés. Au Royaume-
Uni, le mélange de sol sec est relativement courant pour améliorer le sol avec des matières très humides et / ou
organiques. Le mélange humide est moins courant et implique l'introduction d'un coulis fluide avec une
désagrégation simultanée du sol avec un outil de mélange rotatif. Ce document présente les résultats d’essais de
vérification de la résistance menés au Royaume-Uni dans plusieurs projets couvrant différents types de sol et
différents critères de spécification de projet. Certaines conclusions sont tirées en ce qui concerne les facteurs
influant sur la progression de la résistance et sur la compréhension des mécanismes du processus de mélange. Le
temps de mélange par unité de volume de matériau mélangé est identifié comme un paramètre important pour le
mélange en masse. En outre, la modélisation discrète éléments s’est révélée prometteuse pour comprendre les
mécanismes du mélange en colonne profonde.
Keywords: deep soil mixing; mass mixing; soft ground; ground improvement; discrete element modelling
1 INTRODUCTION
Ground improvement technologies are used
extensively in the civil engineering and building
industries to engender higher strength, lower
compressibility or improvement of other
engineering properties into native soils for the
purposes of accommodating greater load or
achieving a greater level of serviceability for a
structure than would have otherwise been
possible. Success of these methods are prevalent
in Japan, the United States of America,
Scandinavia, Great Britain and Ireland
(Munfakh, 1997; Terashi & Tanaka, 1981; Hebib
& Farrell, 2004) having been pioneered initially
and independently in Japan and Scandinavia.
Page 2
B.3 - Ground reinforcement and ground improvement
ECSMGE-2019 – Proceedings 2 IGS
In extensive and deep deposits of soft ground,
deep soil mixing, traditionally encompassing the
mechanical agitation of ground with the addition
of a cementitious or lime binder, is commonplace
as an improvement method. The chemical
processes of binder introduction, (i.e. hydration
& subsequent production of primary & secondary
cementitious by-products, ion exchange &
flocculation, pozzolanic reaction and
carbonation), are well understood with well-
defined relationships between the volume of
binder introduced and the strength and/or
stiffness increase (for a given binder type or
blend). However, the mechanics of the mixing
processes are not well understood.
1.1 Soil mixed columns
Soil mixed columns find particular application in
the treatment of deep deposits of poor materials.
Typically the columns are combined with a soil
mixed load transfer platform to provide a
working formation of high bearing capacity.
Other applications include settlement reducing
techniques and cut-off walls.
Figure 1 shows a proprietary rig-mounted wet
soil mixing system known colloqially as Turbojet
(developed by Trevi Soilmec) which will be a
focus of this paper. The system involves
penetration of a mixing tool at high revolutions
per minute in conjunction with the introduction
of liquid grout under high pressure (typically in
excess of 250 bar). A typical mixing tool is
shown in Figure 2. The combination of the high
number of blade rotations per unit depth and the
disaggregation engendered by the grout under
high pressure results in complete destructuring of
the native soil and, thus, high quality mixed soil.
The system is considered to be a hydrid of
mechanical soil mixing and jet grouting
technologies – its primary advantages being the
facility to work effectively in a broader range of
soil parameters, both granular & cohesive,
including high plasticity clays and high
production rates. Wet soil mixing techniques are
sub-optimal for soils of natural moisture content
in excess of 100% where the native water content
diminishes the effect of grout addition. The
governing construction parameter is the blade
rotation number (BRN), defined in EN
14679:2005.
Figure 1. Turbojet system for deep soil mixing (cour-
tesy of Ground Developments Ltd)
Figure 2. Turbojet mixing tool (courtesy of Ground
Developments Ltd)
Page 3
Some observations on the design and construction of wet soil mixing in the UK
IGS 3 ECSMGE-2019 - Proceedings
1.2 Mass mixing
Mass mixing (see Figure 3) involves
disaggregation of the soil using an excavator-
mounted rotavating tool (see Figure 4). In
general, the principle is the same as deep columns
however mixing usually takes place within
discrete “cells” and depth is limited up to 5-7m
depending on the application and native soil.
The rotavating tool spins at high revolutions
per minute (in the order of 80-90rpm) and grout
is injected under medium pressure. The result is
a completely fluidised cell and mixing time per
cell is conjectured to be critical to homogenising
the material and engendering the required
strength.
Figure 3. Mass mixing system (courtesy of Ground
Developments Ltd)
Figure 4. Mass mixing tool (courtesy of Ground De-
velopments Ltd)
2 BACKGROUND TO PROJECTS
The projects which are the subject of this paper
cover northern England and Scotland as shown in
Figure 5. Table 1 compares and contrasts the
projects in terms of ground conditions and soil
mixing types. In all cases, groundwater was near
(within 1m) commencement level.
Figure 5. UK project locations
Project
location
Indicative soil type Mixing type /
depth
Walney Tidal flats, very soft,
sensitive CLAY
Deep columns
up to 25m deep
Dundee Infilled quarry,
uncontrolled granular
fill & waste (ash)
Deep columns
up to 17m deep
Airdrie Granular Made
Ground with PEAT
lenses up to 1.5m
thick
Mass mixing
up to 4m deep
Stonehouse Very soft, sensitive
CLAY
Mass mixing
up to 5m deep
Aberdeen Very soft SILT with
marine influence
Mass mixing
up to 6m deep
Table 1. Summary of project conditions
Page 4
B.3 - Ground reinforcement and ground improvement
ECSMGE-2019 – Proceedings 4 IGS
3 OBSERVATIONS
3.1 Deep soil mixing
While the focus of this paper is on the laboratory
strength testing for routine quality control of soil
mixing, it is important to note that laboratory tests
should be augmented by visual-manual
inspection where appropriate. This can be
undertaken by exposing mixed material (see
Figure 6 which shows the head of a soil mixed
column exposed) and through extraction of rotary
cores post-construction (see Figure 7).
Figure 6. Exposed soil mixed column (courtesy of
Ground Developments Ltd)
Figure 7. Partial rotary core through soil mix col-
umn (courtesy of Ground Developments Ltd)
3.1.1 Density
In routine strength testing, undertaken as
unconfined compressive strength testing of
manufactured cubes of samples extracted from
site, density of the sample is normally measured.
Observations have not demonstrated any reliable
correlation of strength and sample density. In
Figures 6 & 7 below, strength and density are
compared for 14-day and 28-day cured samples
from the Walney and Dundee sites.
The Dundee data has inherently lower density
owing to the nature of the native mixed soil,
being of granular, ash & waste composition and
this is reflected in the mixed material. However,
this has not affected the overall strength gain.
Both sets of results express similar strengths
(both sites had equal target design strengths).
Figure 8. 14-day strength-density plots
Figure 9. 28-day strength-density plots
Page 5
Some observations on the design and construction of wet soil mixing in the UK
IGS 5 ECSMGE-2019 - Proceedings
3.1.2 Laboratory strength
Histograms are presented in Figures 8 & 9 for the
strength measurement from the Dundee &
Walney sites for 14-day and 28-day strengths
respectively. The results indicate a normal
distribution but with a hint of right skew on the
7-day results. Standard deviations for the 14-day
& 28-day results were 0.34MN/m2 & 0.25MN/m2
respectively – the target 28-day strength was
1.0MN/m2.
The Dundee (predominantly granular) data
cumulatively expressed more rapid strength gain.
However, at 28 days the Walney (predominantly
cohesive) data had exceeded the Dundee strength
cumulatively.
Figure 10. Histogram of 14-day strength test results
Figure 11. Histogram of 28-day strength test results
The mean 14-, 28- & 48-day strengths are
presented in Figure 12 below. The data was found
to fit well with the relationship of Åhnberg
(2006) as follows:
𝑞𝑡𝑞28
≈ 0.3 ∙ ln 𝑡
Where: qt is the strength at time t
q28 is the strength at 28-days
However it should be noted that the sample size
for 48-days is comparably small. For statistical
completeness, the associated box plots for the
Dundee & Walney strength data are presented in
Figures 13 & 14 below respectively.
Figure 12. Strength gain of mixed material from col-
umn samples
Figure 13. Box plot of strength results for Dundee
Page 6
B.3 - Ground reinforcement and ground improvement
ECSMGE-2019 – Proceedings 6 IGS
Figure 14. Box plot of strength for Walney
3.2 Mass mixing
The mass mixing system is a versatile tool with
applications in both permanent & temporary
works. The range of projects considered here
(Airdrie, Stonehouse & Aberdeen) included
bearing capacity improvement for a retaining
wall, car-park and large silo tanks. Other
applications include stabilisation of deep
extremely soft soil for workability, cut-off walls
for groundwater & ground gas and improvement
of passive soils in cofferdams for temporary
excavation support.
The design of the rotavating head itself is
specific to soil type. Figure 15 below shows a
head more suited to cohesive soils where as the
head shown in Figure 4 is more suited to granular
materials.
Figure 15. Alternative mixing head (courtesy of
Ground Developments Ltd)
3.2.1 Mixing time
Experience has shown that mixing time is critical
to the success of mass mixing procedures.
Specifically, mixing time per unit volume of
material in each cell is conjectured to be a critical
measure to ensure adequate strength gain and
homogenity of the mixed material.
In the mass mixing projects discussed here, the
veracity of this perception is examined from full-
scale project data. Mixing time is plotted against
strength gain in Figures 16 & 17 below for 7/14-
day and 28-day strength measurements.
However, given that the mix design is markedly
different for the Aberdeen project, these results
have been normalised by q* (considered as the
mean result plus one-half of a standard deviation
from the mean as an upper bound of the results)
and presented in Figure 18.
Figure 16. 7-day to 14-day strength & mixing time
results for mass mixed sites
Figure 17. 28-day strength & mixing time results for
mass mixed sites
Page 7
Some observations on the design and construction of wet soil mixing in the UK
IGS 7 ECSMGE-2019 - Proceedings
The normalised plot shows that mixing time
potentially has an optimal minimum of between
1.5 and 1.75 minutes. In statistical terms, the
sample range is too narrow to propose a
reasonable model but there is evidence of a peak
(and thus optimal) mixing time where strength
gain is maximised.
The data strongly indicates that mixing time
less than 1.0 minutes has a detrimental effect on
strength gain. Equally, there is evidence of a
tapering off of strength gain with mixing time in
excess of 1.5 to 1.75 minutes.
These observations are very important in terms
of developing project specifications where wet
mass mixing is proposed. Mixing time per unit
volume should be considered as a primary control
parameter. Soil mixing does not currently have a
standardised specification in the UK. Typically,
projects where soil mixing is proposed will have
a bespoke specification identifying geotechnical
properties for the mixed material. EN
14679:2005 is normally proposed as a guiding
code of practice. While the BRN is a useful
control parameter for deep columns, there is no
equivalent control parameter for mass mixing
appartus. Mixing time per unit volume appears to
be a purposeful measure to this end.
Further work and data collection over a wider
range of mixing times is needed to form a firmer
view on the precise relationship between mixing
time and strength gain.
Figure 18. Relationship of mixing time and strength
development
4 FUTURE WORK & RESEARCH
4.1 Data development
The observations presented here are based on a
limited project set, though reflect a broad range
of soil types and project applications. It is hoped
that the data will be continuously augmented in
order to develop understanding of the
mechanisms of strength development in mixed
material.
The mechanisms controlling strength gain in
deep columns are better understood (e.g. BRE,
2002) however, the mechanisms controlling the
success of mass mixing applications are less well
understood. A broader database of soil types,
project applications and field & laboratory
measurements is needed to develop firm theories
and mathematical models of the governing
mechanics.
4.2 DEM modelling
The excellent work of Larsson (2003) outlines the
state-of-the-art in the understanding of mixing
processes for soil mixing applications.
Some recent work by O’Brien (2018) studying
the mixing mechanisms using discrete element
modelling (DEM) has provided some useful
qualitative insight into the mechanisms
governing the BRN. The DEM model (see Figure
19) uses particle trace to track individual
elements during the mixing process and chart the
particle behaviour (see Figure 20). The
qualitative conclusions of the study outline how
higher BRNs cause cyclical migration of the
particles in the plane of rotation about the tool but
also laterally paerpendicular to the tool, thus
inducing a better quality of mix.
Expansion of such studies is required to
develop a numerical framework for both the
qualitative & quantitative understanding of
mixing mechanisms and to supplement the
empirical observations derived from field and
laboratory data as highlighted in this paper.
Page 8
B.3 - Ground reinforcement and ground improvement
ECSMGE-2019 – Proceedings 8 IGS
Figure 19. Particle trace during soil mixing using
deep column mixing tool
Figure 20. Graphical representation of particle
movement during mixing
5 CONCLUSIONS
The results of field & laboratory observations
have been presented for both deep column mixing
and mass mixing for UK-based projects using
wet (grout-based) techniques. For deep column
mixing, no discernible correlation was found
between sample density & strength. In addition,
the development of strength was observed to
follow the time-based progression of Åhnberg
(2006) well irrespective of whether the native soil
was predominantly granular or cohesive.
In terms of mass mixing, mixing time per unit
volume has been shown to potentially be a
governing parameter in the strength gain
progression. Field and laboratory observations
suggest that there is an optimal mixing time of
between 1.5 and 1.75 minutes per cubic metre of
mixed material. It is suggested that this parameter
should be used as a control parameter, analogous
to the blade rotation number used in deep column
mixing, for projects involving mass mixing.
It is noted that there is much scope for further
studies particularly around understanding of the
mixing mechanisms. Discrete element modelling
has shown promise in potentially setting up a
numerical framework but further development of
field & laboratory databases is also advocated.
6 ACKNOWLEDGEMENTS
The author would like to thank colleagues at
Ground Developments Ltd for data, images and
thoughtful comments & practical insights in the
preparation fo this paper.
7 REFERENCES
Munfakh, G.A. (1997) “Ground improvement
engineering – the state of the US practice: part
1. Methods”. Proc. Institution of Civil
Engineers Ground Engineering, vol. 1 pp. 193-
214
Terashi, M. and Tanaka, H. (1981) “Ground
improved by Deep Mixing Methods”, Proc.
10th ICSMFE, vol. 3, pp. 777-780
Hebib, S. and Farrell, E.R. (2004) “Stabilisation
of Irish Soils”. Inst. of Engineers of Ireland
BSI (2005) EN 14679:2005 “Execution of
Special Geotechnical Works - Deep Mixing”
British Research Establishment (2002) “Design
Guide Soft Soil Stabilisation CT97-0351
(EuroSoilStab)”
Åhnberg, H. (2006) “Strength of Stabilised Soils:
A Laboratory Study on Clays and Organic Soils
Stabilised with Different Types of Binder”
Swedish Deep Stabilisation Research Centre
Report 16
O’Brien, A. (2018) “A simplified parametric
study of particle trace in soil mixing simulations
using discrete element modelling” MSc. Thesis
School of Mathematical Sciences, University
College Cork, Ireland